Dear Simon,
your comments were also very helpful for me as the normalisation by means of the EPI template do not seem to work very well.
I just have a brief question. You mentioned that in the Realignment the reslicing should only be done for the mean only. Does this mean that the other images are not realigned and only normalised into the MNI space? Or are the realignment parameters stored in the headers of the files and considered in the reslicing during normalisation?
Thanks a lot for sharing your expertise with us!
Best wishes,
Roland
-----Ursprüngliche Nachricht-----
Von: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] Im Auftrag von Simon B. Eickhoff
Gesendet: Dienstag, 09. Jänner 2007 18:34
An: [log in to unmask]
Betreff: Re: [SPM] AW: [SPM] manual (preprocessing)
Dear Markus
> dear simon.
> your answer helps a lot, but generates further questions (i am very
> sorry about that):
>
> >Conseqently the only time, where a reslicement is necessary is the
> spatial normalisation, which is non-linear and cannot be stored
> with the images.
> that means, only in normalization and smoothing i need to reslice?
Yes, that was my point.
> >Realign the EPIs and create a mean EPI
> > (Realign & Reslice, Reslice: "mean only")
> ok, this is what i have done with my images in spm2.
And you can do the same in SPM5
> i am absolutely not familiar with the segmentation algorithm in
> spm5 and why i should avoid to use a epi template.
I would suggest avoiding the EPI template, as it is not strictly speaking in the MNI reference space (john, corret me, if I'm wrong). You might, however, want to register your data as good as possible into this space in order to use the same frame of reference as other functional and anatomical studies (OK, this assumes, most people use the MNI templates).
> what do i need the t1 image for because you normalize epi with epi, dont you? if i still need
> the t1 image, should it be skull stripped?
You don't strictly need the T1 image, if you normalise your EPI. But it is always useful to carry it along so that you can check your functional data with respect to the individual anatomy.
> >>- Apply the derived normalisation-parameters to all your images
> (meanEPI, individual EPIs, coregistered T1) how can i do this?
Normalise -> wirite only -> select the _sn.mat files created during segmentation or normalisation, the images you want to normalise and the desired voxel size after normalisation (you can sick to the defaults here)
> normally i do (after realignement):
> - to coregister the realigned epis (sources and other images) to the t1 (target)
> - afterwards to normalize the t1 with the t1 template
> - write the transformations onto all epis.
Thats possible as well. However, your coregistration wil take longer (as you write hundreds of images) and your normalisation will potentially be flawed by the non-linear differences between T1 and EPI.
> i do not like to bother you with my questions to much, but i need more
> information about that. is there a text which refers to your procedure
> of segmentation.
Ashburner et al., 2005 "Unified segmentation"
Hope this helps
Simon
>
> thanks a lot.
>
> markus
>
>
-----------------------------------------
Simon Eickhoff
Institut for Medicine (IME); Research Center Jülich
Leo-Brandt-Str. 5; 52425 Juelich, Germany
Phone + 49 2461-61-5219 / Fax + 49 2461-61-2820
e-mail: [log in to unmask] / [log in to unmask]
|